SUMMARY OF WORK: PURPOSE: Eleven years ago, I proposed a new model of the immune system (the Danger model) based on the assumption that the immune system's function is to discriminate between dangerous and harmless things rather than self and non-self. Because this model has tremendous implications for such subjects as cancer, transplantation, neonatal and adult vaccines, parasitology, and autoimmunity, we have been testing its basic premises and its applicability in several areas. RESULTS FROM THE PAST YEAR: 1) NEONATAL IMMUNITY: neonatal mice respond to a vaccine, even if they have received passive antibody from their mothers. The immunity in babies from immunized mothers lasts for at least a year, and does not fade any more rapidly than immunity in babies from control mothers. Thus, in mice, maternal immunity does not seem to be inhibitory 2) TOLERANCE: mice and sheep immunized against their own milk proteins make T cell proliferative responses, and several classes of milk-protein specific antibody. However, they do not have autoimmune disease when they are bred and begin to lactate. Thus the immune system has a mechanism to instigate tolerance to proteins that appear late in life, even after immunization against those proteins 3) T CELLS: CD4 helper T cells are not only able to reject tumors in the absence of CD8 killers, they are better at it than CD8 cells. Against six out of six tumors, from five different tissues, CD4 effectors were more potent than CD8s. 4) DENDRITIC CELLS: the commonly held view that dendritic cells secrete IL-12p70 upon LPS stimulation, and thus directly control the initiation of Th1 responses, is incorrect. We found that it is the p40 subunit, not p70 that is made. IL12 p70 is only produced by dendritic cells if they are stimulated to do so by previously activated T cells.